Bug 88378

Summary: Remove JSObject::m_inheritorID
Product: WebKit Reporter: Gavin Barraclough <barraclough>
Component: JavaScriptCoreAssignee: Mark Hahnenberg <mhahnenberg>
Status: RESOLVED FIXED    
Severity: Normal CC: barraclough, fpizlo, mhahnenberg, mitz, ossy, zarvai
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Bug Depends on: 92338    
Bug Blocks:    
Attachments:
Description Flags
Fix
none
Patch fpizlo: review+

Gavin Barraclough
Reported 2012-06-05 17:12:51 PDT
This is rarely used, and not performance critical (the commonly accessed copy is cached on JSFunction), and most objects don't need an inheritorID (this value is only used if the object is used as a prototype). Instead use a private named value in the object's property storage.
Attachments
Fix (9.52 KB, patch)
2012-06-05 17:16 PDT, Gavin Barraclough 		no flags
DetailsFormatted DiffDiff
Patch (9.27 KB, patch)
2012-07-25 17:05 PDT, Mark Hahnenberg 		fpizlo: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Gavin Barraclough
Comment 1 2012-06-05 17:16:36 PDT
Created attachment 145900 [details] Fix
Geoffrey Garen
Comment 2 2012-06-05 17:19:28 PDT
Comment on attachment 145900 [details] Fix r=me
Gavin Barraclough
Comment 3 2012-06-05 22:08:55 PDT
Fixed in r119556
mitz
Comment 4 2012-06-06 00:33:31 PDT
(In reply to comment #3) > Fixed in r119556 This broke the 32-bit build, as far as I can tell because it made the m_inlineStorage data member in JSNonFinalObject not 4-byte aligned, causing the assertion in finishCreation() to fail.
mitz
Comment 5 2012-06-06 00:37:33 PDT
I didn’t know how to fix the build, so I reverted this change in <http://trac.webkit.org/r119568>.
mitz
Comment 6 2012-06-06 00:38:09 PDT
Comment on attachment 145900 [details] Fix Changing to r- because it breaks the 32-bit build.
Gavin Barraclough
Comment 7 2012-06-07 21:29:58 PDT
Re-landed in r119795
Zoltan Arvai
Comment 8 2012-06-08 01:22:44 PDT
(In reply to comment #7) > Re-landed in r119795 Layout test fails on Qt, GTK, EFL and Lion after r119795: jquery/core.html
Gavin Barraclough
Comment 9 2012-06-08 01:55:23 PDT
Ugh. I'll revert in the morning if no-one else has done so, & debug. cheers for reporting this.
Gavin Barraclough
Comment 10 2012-06-08 14:31:41 PDT
Rolled back out in 119865. :'-(
Mark Hahnenberg
Comment 11 2012-07-25 17:05:24 PDT
Created attachment 154496 [details] Patch
Mark Hahnenberg
Comment 12 2012-07-25 17:13:23 PDT
Committed r123682: <http://trac.webkit.org/changeset/123682>
Csaba Osztrogonác
Comment 13 2012-07-25 23:26:49 PDT
(In reply to comment #12) > Committed r123682: <http://trac.webkit.org/changeset/123682> It made many tests crash on 32 bit platforms. Could you check it, please? new bug report for this regression - https://bugs.webkit.org/show_bug.cgi?id=92338
Csaba Osztrogonác
Comment 14 2012-07-25 23:34:54 PDT
Fixed by http://trac.webkit.org/changeset/123708. Thanks.
Note You need to log in before you can comment on or make changes to this bug.